Visit the Zuken web site
Click on the advert above to visit the company web site

Product category: Electronics Manufacturing Quality Assurance
News Release from: Machine Vision Products | Subject: Ericsson Lynchburg
Edited by the Electronicstalk Editorial Team on 18 October 2001

Implementing dynamic process control

Request your FREE weekly copy of the Electronicstalk email newsletter. News about Electronics Manufacturing Quality Assurance and more every issue. Click here for details.

Engineers at an Ericsson basestation plant have used existing information from AOI systems to achieve greater throughput, lower defect rates and complete closed loop process control.

The history of PCB assembly is highlighted by the continued automation that now drives high-speed electronics manufacturing Step by step, each segment of the process, from screen printing to placement to reflow, test and now into final assembly, has been made faster by what can only be described as mind-blowing advances in machines and software

But the landscape is changing as the industry matures.

Today's price-sensitive market realities beg two very important questions.

Can next generation automation continue to substantially lower manufacturing costs and push profitability? If not, what is the next development that will drive EMS and OEM companies? More than esoteric queries suggested by ivory tower theorists, the questions are of critical concern for today's manufacturers, as well as for industry organisations working to develop the roadmaps that will, it is hoped, guarantee the strength and health of the electronics industry.

According to one such organisation, the National Electronics Manufacturing Initiative (NEMI), the key to future profitability has shifted from automation to sound business practices.

As quoted in Electronic News (15th January 2001) NEMI's president, Jim McElroy, believes, "The prime source for continuing cost reduction has migrated from manufacturing efficiency to improved supply chain management...Industry has optimised manufacturing efficiencies, and continued productivity gains in materials conversion will only be incremental...".

McElroy's analysis comes as no surprise to high-end manufacturers or international EMS companies.

By themselves, "incremental" increases in profitability are not enough to overcome downward price pressures.

How then do you push profits and reduce costs beyond NEMI's better business practices? For Ericsson in Lynchburg, Virginia (Plant 52 to be exact - and now part of SCI), much of this solution involves taking advantage of digital technologies and the full capabilities of automated systems.

The key word is "full".

Ericsson engineers wanted to use existing information gathered by AOI systems to achieve greater throughput, lower defect rates and complete closed loop process control of their basestation SMT line.

In part, Ericsson's vision of improved line efficiency runs counter to the industry's traditional focus: that is, a narrow vision focused on individual machines.

The reason for this limited focus is understandable: Automation was put in place one process at a time - the concept of total line efficiency, from pre- to post-reflow, has taken a back seat to the simple desire to optimise the line one machine at a time.

But such tunnel vision can no longer be the order of the day.

Along with best management practices, profitability and quality depend upon quantifiable line optimisation, independent of equipment brands or models, increased throughput and real time SPC across the entire line.

With these procedures as standard industry practice, only then can it be claimed that manufacturing efficiency has reached its apex.

Ericsson (now SCI) Lynchburg operates six high-speed SMT lines, each with varying degrees of automation.

Under the direction of Brian M Barley, Technical Leader, SMD Engineering, one of the lines is dedicated to manufacturing base stations - a five product series of transmitters/receivers.

Each product requires a sophisticated 8-layer single-sided board using standard SMT and with large through-hole connectors.

The connectors are placed as a pin and paste operation.

Each board holds three connectors - two larger ones (96-pin connector and a dual coaxial connector) and one single coax connector.

The single connector is placed using a Universal GSM machine, the two larger ones are hand inserted prior to reflow - an operation that is still required because the parts vendor does not have machine acceptable packaging.

The board itself includes 0603 packages and a dozen or so 20mil packages (QFPs and TSOPs), which are supplied via tape and reel and placed with the GSM.

The line is setup to essentially run a lot size of one.

From product to product, the boards vary just slightly.

There is no changeover on the line - other than reselecting specific programs.

As currently configured, the line consists of an MPM screen printer, an MVP AutoInspector 1820 AOI machine for paste inspection, a Fuji CP642, a second CP642, a Fuji CP6, a Fuji IP-III, a Universal GSM, a Speedline Electrovert oven and another MVP 1820 for post-reflow inspection.

Four hundred boards are produced per shift, and the facility runs three shifts a day, five days a week.

Two years ago, Brain Barley and Ericsson engineers went searching for ways to improve yield and reduce defects - and did so by looking at the entire line and not just individual machines.

"It seemed to us that there was a void - a lack of looking at the manufacturing process as a big picture, as an SMT line", explains Barley.

"We wanted to integrate every piece of the process and provide closed loop feedback across all machines and all vendors using a single common platform.

At the time, there were ways of achieving this goal, but only through the use of several systems, which together were cumbersome.

We were looking for a single system, a single point of contact".

Ericsson also wanted a common feeder verification system across all machine platforms.

Historically, the basestation line used the Fuji Handy Terminal (HT) - an optional Fuji factory-based system that offers feeder verification.

"But the HT only does part number verification and has no way of handling operator IDs, slot verifications or feeder IDs.

And it only runs on Fuji equipment", says Barley.

"We knew we were adding a Universal GSM placement machine so we needed something more".

For Ericsson, the solution began with Machine Vision Product's (MVP) AutoInspector 1820 AOI system and led to MVP's CIMCIS and Dynamic Process Control technology.

Using the standard defect data and variable measurements that today's sophisticated AOI systems can generate, it is possible to create a real time inspection/defect system that immediately alerts line operators whenever any process exceeds preset limits.

This is, in essence, real-time SPC, a system that continually monitors line performance, detects problems with each board, measures pick and place performance, measures feeder and nozzle performance and stringently controls process variability to optimise performance, throughput, first pass yields and overall quality.

The two keys to such a system are fast, flexible AOI, linked to a series of line controllers and an intelligent network that maintains AOI data and quickly translates the circuit reference data into specific machine, feeder and nozzle information.

Using data generated by the AutoInspector 1820, MVP developed a system that links AOI to a fully scaleable, plug and play hardware/software system, which uses standard AOI information to monitor and dramatically improve the performance of SMT production lines.

Dubbed MVP CIMCIS, the system uses small, relatively inexpensive data terminals to collect and store production information and automatically calculate SEMI E-10 Metrics - measurements such as best and average cycle times; run, blocked, starved and down times and the top 10 feeder and nozzle problems.

The line controller and data server use an RS485 network to collect data.

The network allows the transfer of data from the screen printer, the pick and place machine, the reflow oven and the AOI system - regardless of brand.

The result of this "AOI-SPC" combination is dynamic process control, a system that goes beyond SPC - adding audible alerts and, via a visual display panel, proactively suggesting to the operator what corrective actions should be taken.

By calculating best and average cycle times, MVP CIMCIS lets operators know how much system utilisation is actually being achieved.

It's not unusual for the addition of dynamic process control to improve pick and place utilisation anywhere from 5 to 70% and beyond.

And the collected data can be used to track faulty nozzles and feeders back to the manufacturer, track faulty components to a specific reel, and quickly optimise the line to meet a changing product mix.

In part, Barley was led to MVP CIMCIS because other Ericsson lines at Lynchburg, particularly cellular phone manufacturing, were already using MVP 1820s for high speed AOI.

And the base station line itself had two 1820s for pre- and post-reflow inspection.

Starting two years ago with CIMCIS terminals and a line controller in place on Fuji placement machines - at that time CP4-3 machines - Ericsson began the process of collecting critical data and proactively replacing faulty feeders and nozzles as they were identified.

"In so doing, we also were able to get away from a paper system of logging our feeder changes", says Barley.

"With CIMCIS everything is database logged.

If we have any issues about what part was loaded on what machine, when and by whom, we go to the database and pull the specific information we need".

Starting with two terminals, Ericsson was able to use CIMCIS data to immediately track nozzle and feeder errors and downtime.

"The base system worked from the start, and gave me information about how well each Fuji machine was operating.

Quickly, we realised that we did indeed have feeder problems that were going undetected before MVP CIMCIS - particularly on the CP4-3 machine, which used an older feeder design.

CIMCIS found a lot of feeders that were generating errors and mispicks, which we weeded off the machine.

With this immediate feedback we could see what feeders were giving us problems.

These improvements were literally 'out of the box' and represented the first big gains we made with CIMCIS.

Throughput improved because we were picking more parts accurately; we weren't losing time due to mispicks, and the machine was not stopping due to multiple mispicks that require the operator to change a feeder".

During this initial testing phase of MVP CIMCIS, hard numbers were not kept.

But as the Ericsson plan evolved and more CIMCIS terminals were added, Barley was able to quantify his numbers.

Today there are CIMCIS terminals on both Fuji CP642s, the CP6, the IP-III, the Universal GSM and the post-reflow MVP 1820.

"Under the new line configuration we have been fully operational with all machines for five months, and we have been tracking results.

From a quality standpoint our first pass yields have increased by 5%.

We also track defect as PPM and we break these numbers down by solder defects and placement defects.

CIMCIS as it is now configured on our line is directly related to placement defects, and during Phase I placement defects were reduced on average across the five products by 50ppm".

Barley plans on tracking solder defects once a CIMCIS terminal is added to the MVP paste inspector - something planned for phase II implementation.

Ericsson and Barley played a large role in developing the closed loop feedback capabilities of CIMCIS.

As part of the ideal solution, Barley wanted to take the CIMCIS collected information and feed it back to the point in the line where the defect was created, and do so in a timely manner.

What Ericsson was trying to avoid is the still very typical routine of having an operator or engineer look at an error report hours after the fact, and only then affect the necessary change.

Ericsson wanted immediate or close to immediate verification of the problem and an accurate pinpointing of the source.

Working with MVP CIMCIS hardware and software engineers on site, the information loop was closed.

"We have gone from a matter of hours before we could take corrective action to a matter of minutes", says Barley.

"The only time frame we have now is the time it takes the PCB to get to the MVP 1820".

Given the speed of today's SMT lines, waiting hours to correct a problem means building thousands of incorrect boards, and wasting time, money and resources.

Using real time proactive feedback, it is possible to correct problems almost instantly.

The CIMCIS terminal not only alerts operators to potential problems, but also suggests corrective action.

"One of the things that sent us to MVP CIMCIS was the speed of the 1820 AOI machine", says Barley.

"In fact, speed is why the cellphone operations used MVP - as they have very low tack times.

This speed helps with basestations because we are running large, highly populated PCBs, and the faster we can inspect the more time we have at the repair bench.

But we were also attracted by MVP's willingness to work with us.

Not many vendors are eager to do so.

Usually vendors deliver an off the shelf package and say 'take it or leave it.' MVP was not that way - and this attitude is something we liked about the company from the very outset.

And most importantly, we liked the fact that MVP CIMCIS interfaces across multiple machine vendors, offers a common platform and had the potential for closed loop feedback".

Combining AOI data with CIMCIS collection modules placed throughout an SMT production line is changing static SPC into real time SPC.

The result of using automation to its fullest is higher throughput and yield, greater line efficiency and lower overall manufacturing costs.

For Ericsson's complex basestations, the push for greater throughput and integrated line optimisation was solved by linking high speed AOI to "platform agnostic" CIMCIS terminals; thus closing the information loop and creating, in essence, dynamic process control.

If the increases in speed and profitability once brought to the industry by automation are now merely incremental, the next breakthrough may be how wisely we actually use the automation at hand.

It's no longer enough to think that individual machines are working and that the lines themselves "seem" to be optimised.

As Ericsson found out, there's a lot to be gained by collecting data, monitoring each step along the way, and analysing information so that operators can, for the first time, "know" how much automation is actually working and how much remains as mere "potential".

Machine Vision Products: contact details and other news
Email this article to a colleague
Register for the free Electronicstalk email newsletter
Electronicstalk Home Page

Search the Pro-Talk network of sites

Visit the Zuken web site